JPH01284984A - Picture processing method - Google Patents

Abstract

PURPOSE:To attain the exact calculation of the feature quantity of a recessing part by evaluat ing a distance between a straight line, which connects mutually adjacent enveloping points, and a border picture element between both the enveloping points or arranging picture elements between both the enveloping points while the distance with the straight line is evaluated and forming a projecting closure. CONSTITUTION:The border picture element is divided into four quadrants and border picture element data are arranged in a chain order. Then, data, in which an index is given to the border picture element to be the enveloping point, are generated and the inclination of the straight line, which connects the mutually adjacent enveloping points, is obtained. The border picture element between the both enveloping points is successively pursued and the distance between the center of the border picture element and straight line is evaluated. Then, the recessing part is detected. For the calculation of a recessing arc length, the border picture element is successively counted from the first border picture element, which is positioned in the graphic inside of the straight line to connect both the enveloping points. Then, when the border picture element is not the recessing part, a count value is made ineffective and when it is the recessing part, the number of the border picture elements in the recessing part is obtained as the recessing arc length. Thus, the feature quantity of the recessing part can be exactly calculated.

Description

DETAILED DESCRIPTION OF THE INVENTION [Technical Field of the Invention] The present invention relates to an image processing method for determining a waiting amount related to a concave portion of a figure.

[Background of the invention]

For example, the handwritten document "R" shown in Figure 1 is 1! ! When you realize
The decisive factor is the presence of 211 recesses, recess 1 and recess 2, and the presence of one hole. In this sense, the unevenness of a figure is an important feature of the figure. The minimum convex figure that includes a figure is called a convex hull, but the figure obtained by removing the original figure from the convex hull is I! l! Parts and holes are extracted and are important for the analysis of the geometry. Assuming that there is a figure whose envelope points have already been determined, it is possible to generate a figure that is close to the convex hull by sequentially connecting two adjacent envelope points with a straight line. However, the generation of straight lines in digital figures is done by the algorithm As a result, there is a variation of one pixel vertically and horizontally, and it is not necessarily possible to form a convex hull.Also, if you select the pixel closest to the straight line of the analog figure, there is a possibility that a pixel outside the convex hull will be selected. The generation of basic figures such as straight lines is often performed by a drawing processor, but in this case, the generation of straight lines depends on the algorithm of the drawing processor, and there is no guarantee that a correct convex hull can be generated. When a convex hull is generated, the figure obtained by removing the original figure from the convex hull, which includes only concave portions and holes, also becomes inaccurate, which causes problems in subsequent analysis.

In addition, there are concave chord lengths, concave arc lengths, date dials, etc. that indicate the shape of concave parts, but since there is no particularly effective method for convex closure, there is no efficient method for calculating the waiting amount. It didn't exist.

[Purpose of the invention]

The present invention was devised to solve these conventional problems, and relates to an image processing method that can accurately and efficiently calculate the feature amount of a concave portion.

[Summary of the invention]

The 1iii#i processing method according to the present invention arranges boundary pixel data in chain order, generates data in which indexes are given to boundary pixels serving as envelope points, and calculates the slope of a straight line connecting adjacent envelope points. The boundary pixels between both envelope points are sequentially traced from one envelope point to the other, and the distance between the center of the boundary pixel and the straight line is evaluated to detect a concave part. When calculating the arc length, the boundary pixels are counted sequentially from the first boundary pixel located inside the figure from the straight line connecting the two encompassing points, and if it is found that it is not a concave part, the count value is invalidated and When it turns out to be a recess, the boundary within the recess 11] The prime number is the recess arc length, and when calculating the recess arc length, when it is determined that it is a recess, calculate the length of the line segment that connects the starting point and end point of the recess. When calculating the number of recesses, the number of recesses is counted up when it is determined that the recess is a recess.

In another image processing method according to the present invention, the slope of a straight line connecting adjacent envelope points is determined, and pixels located at the shortest distance from this straight line and located inside the figure from one envelope point to another are arranged in order toward the envelope point to form a convex hull, and the figure obtained by subtracting the original figure from this convex hull is a figure with only concave parts, and when calculating the concave chord length and concave arc length,
An index is attached to the pixels arranged between the envelope points, and in a figure consisting of four parts, the pixels with the index are counted as the concave chord length, and the pixels without the index are counted as the concave arc length, and [! l! Regarding the calculation of I, only the shape of the concave portion is labeled.

[Embodiments of the invention]

Next, one embodiment of the image processing method according to the present invention will be described with reference to the drawings.

FIG. 1 shows a figure including recesses A, B, C, D, and E (shown with hatching), and a horizontal rectangle R circumscribing this figure is drawn. This horizontal rectangle R divides the boundary pixels of the figure into four quadrants. In Figure 1, the boundary pixels facing the upper left vertex of the horizontal rectangle are quadrants ■,
The boundary pixel group facing the lower left vertex is quadrant (2), the boundary pixel group facing the lower right vertex is quadrant II+, and the boundary pixel group facing the upper right vertex is quadrant (2).

These quadrants provide useful information for detecting recesses.

A table as shown in FIG. 2 is created in which boundary pixels are arranged in chain order, including information on this quadrant. This table includes information on whether the boundary pixel is an envelope point (in Figure 2, envelope points are indicated with a 0 mark, and non-envelope points are indicated with an x mark), the X coordinate, the X coordinate, and the chain code. It becomes more.

Figures 3 (a) to (d) show the relationship between each quadrant and the chain code. Figure 3 (a) shows the chain code that can be taken by pixels other than the concave portion in quadrant )
shows a chain code that can be taken by pixels other than the concave part in quadrant II, FIG. 3(C) shows a chain code that can be taken by pixels other than the concave part in quadrant I11, and FIG.
d) shows a chain code that can be taken by pixels other than the concave portion in quadrant (■). That is, in quadrant I, when the chain code takes a value other than 4 to 6, it becomes immediately clear that a concave portion exists, and in quadrant I+, when the chain code takes a value other than 6 to 0, it becomes clear that a concave portion exists.
The existence of a recess becomes clear from chain codes other than 0 to 2 in quadrant 1 and from 2 to 4 in quadrant 3.

However, there are cases where it is not possible to determine the presence of a recess using only one chain code, and in such a case, the following process is performed.

In Fig. 4, when the adjacent envelope points PI and P2 are known, the slope of the straight line l connecting PIP2 (1 in Fig. 4 is the slope 5/7 = 0.714, the slope angle θ = 3s, 5 degrees ), the boundary pixels are sequentially tracked from one envelope point, for example, P, toward P2.

Here, the distance d between each boundary pixel and the straight line is given as an integrated value of distance changes Δd based on the chain code of each pixel. In the case of FIG. 4, the relationship between the chain cord and the distance change Δd is as shown in Table 1.

Table 1 In Table 1, the distances of pixels located inside the figure from the straight line are taken as positive, and the distances of pixels located outside of the straight line are taken as negative.

Boundary pixel P b I-Pb between envelope point P + Pa in Fig. 4
Finally, if the integrated value of the distance when tracking from p+* is calculated based on Table 1, it will be as shown in Table 2.

Table 2 As is clear from Table 2, Pb+ and Pb2 are located inside the figure from straight line 1, but the distance from the straight line is 1 m d +
is 0.466 at maximum, so it is generally not determined to be a concave portion, and the distance It d 2 exceeds 1.135, which is 1, for the first time at the boundary pixel Pba, and it is generally determined that it is a concave portion here.

As this boundary pixel is tracked, Pbl to Pb7 are counted, and when it is not a concave part, the count value is invalidated, and when it is a concave part, the previous currant value is valid, and the value immediately before the next envelope point P2 is invalidated. By counting up to the boundary pixels, the concave arc length is calculated.

When it is found that a recess exists between the envelope points PIF2, the distance between the two envelope points is calculated from the X and Y coordinates of the envelope point PIP2, and the concave arc length is calculated. The following four lines are calculated from the concave chord length, concave arc length, and perimeter length.

Headwheel=concave arc length/(perimeter length×concave chord length) In the figure of FIG. 5, the envelope point P1. Boundary pixel Pb between P2
, ~Pb, is present, but it is 0.2 degrees as in Table 2 and 1 above.

Table 3 As is clear from Table 3, the distance for Pb4 is “1.
exceeds Pb3, and once becomes less than "l" at Pb3, Pb
4 exceeds "1". In this case P b l−P b
Let s be one depression, or pb, -Pb3 and II
, Pba-Pbste 111, a total of 2al concave portions, and the concave arc length, concave chord length, mouth wheel, etc. are calculated.

Table 1 is a distance change calculation table for quadrant I, or Table 4 is for quadrants 1 - . Incidentally, the inclination angle θ of the straight line is set as shown in FIG.

Table 4 In the above determination of a recessed part, the distance ad≧1 was used as a condition for the presence of a recessed part, but an appropriate condition such as d>2 may be set in consideration of noise and the like.

When performing the above processing, read the table in Figure 2 one after another and read the product of distance changes) Ell! The conditions shown in Fig. 3 are judged while taking the following values. Then, the integrated value reaches a predetermined value, or the 311th! When the condition of i is not met, it is determined that there is a recess. By counting the boundary pixels while tracking these boundary pixels, the concave arc length can be determined at the same time, and by calculating the distance of the envelope point after confirming the existence of the concave part, the concave chord length can be calculated.

A conceivable method for calculating the feature amount related to the concave portion is to form a convex hull and then subtract the original figure from this convex hull to generate a figure of only the concave portion. However, as mentioned above, there is no guarantee that an accurate convex hull will be obtained by simply connecting the envelope points with straight lines. ing. The method is as follows.

7th +1! ! are the envelope points a, b and the boundary pixels C3~
c9, and when forming a convex hull from this, select the pixel closest to the straight line l defined by the envelope points a and b and inside the figure from the straight line l, from the envelope point a to the envelope point. I'm going to line it up. The order of the envelope points can be either counterclockwise around the figure or clockwise, but the counterclockwise case will be explained here.

First, the chain code in the direction from envelope point a to envelope point S is either "4" or "5J", and the pixel d in the direction of chain code "4" has the center of the pixel closest to the straight line ρ. However, this pixel d° is located outside the figure from the straight line 1. Therefore, the pixel d1 with the chain code "5" is placed adjacent to the envelope point a. From now on, pixels d2 to d whose center is closest to the straight line ρ and whose center is located inside the figure from the straight line l
Arrange 8 toward the envelope point. As a result, the smallest convex figure including the figure is formed.

Here, the distance between the pixels d1 to dS and the straight line ρ is 1.4
Calculated based on.

If a certain index, such as a specific density, is given to the pixels d1 to d8 when forming a convex hull in this way, when a figure with only concave parts is formed, the number of boundary pixels other than the pixel with that index will be The concave arc length is the concave arc length, and the number of pixels with the index is the concave chord length.

〔Effect of the invention〕

As mentioned above, the image processing method according to the present invention evaluates the distance between the straight line connecting adjacent envelope points and the boundary pixel between both envelope points, or evaluates the distance between both envelope points while evaluating the distance to the straight line. Since the pixels are arranged to form a convex hull, it is possible to accurately calculate the feature amount of the concave portion. Further, the processing procedure is to sequentially track the boundary pixels or surroundings of the figure, and since sequential processing can be performed, the processing efficiency is also high.

[Brief explanation of the drawing]

FIG. 1 shows a shape with a concave portion. Figure 2 is a table listing boundary pixel data of the figure to be processed, Figure 3 is a conceptual diagram showing the chain code that is the condition for determining a concave part, and Figure 4 is a table showing the boundary pixels of a figure that includes a concave part. FIG. 5 is a conceptual diagram showing another example of a shape including a concave portion, and FIG.
The figure shows the slope of the straight line connecting the envelope points. The conceptual diagram, Figure 7, shows the process of forming a convex hull, tI! It is l. A-E...Concavity, ■~■...Quadrant, pHP21 a + b """ Envelope point, P b+-P
bs, c+-cs""" boundary pixel, dl-ds...
...pixel.

Claims (7)

[Claims] (1) Arrange the boundary pixel data in chain order, generate data that gives an index to the boundary pixel that becomes the envelope point, find the slope of the straight line connecting adjacent envelope points, and calculate the boundary pixel data between both envelope points. In an image processing method that detects a concavity by sequentially tracing the image from one envelope point to the other envelope point and evaluating the distance between the center of a boundary pixel and the straight line, The boundary pixels are counted sequentially from the first boundary pixel located in the concave area, and when it is found that it is not a concave part, the count value is invalidated, and when it is found that it is a concave part, the concave arc long pot is calculated from the number of boundary pixels in the concave part. Image processing method. (2) When it turns out to be a recess, the starting point of the recess;
2. The image processing method according to claim 1, wherein the length of the line segment connecting the end points is calculated and this length is used as the concave chord length. (3) The image processing method according to claim 1, which counts up the number of recesses when it is determined that they are recesses, and thereby calculates the number of recesses. (4) Arrange the boundary pixel data in chain order, generate data that gives an index to the boundary pixel that becomes the envelope point, find the slope of the straight line connecting adjacent envelope points, Image processing that sequentially arranges pixels located inside a figure from a straight line from one enveloping point to the other to form a convex hull, and then subtracts the original figure from this convex hull to create a figure that consists only of concave parts. Method. (5) The image processing method according to claim 4, wherein the pixels arranged between the envelope points are provided with indices, and the concave chord length is determined by counting the pixels with the indices in the figure consisting of the concave portion. (6) The image processing method according to claim 4, in which the concave arc length is determined by attaching an index to the pixels arranged between the envelope points and counting the pixels to which no index is attached in a figure consisting of a concave part. . (7) The image processing method according to claim 4, characterized in that the number of depressions is determined by labeling a figure containing only the depressions.